{
  "schema": "hangingcontext.desk/1",
  "type": "story",
  "id": "550",
  "slug": "open-weight-agentic-race",
  "edition": "2026-10-24T12",
  "url": "https://hangingcontext.com/stories/2026-10-24/open-weight-agentic-race/",
  "newsroom": "artificial-intelligence",
  "beat": "models-labs",
  "newsroom_title": "Artificial Intelligence",
  "angle": "Open-weight models now compete on agentic benchmarks, not just chat scores",
  "headline": "Open-weight models are now fighting on agentic benchmarks, and the gap is closing fast",
  "dek": "Kimi K3.1 Pro and MiniMax M3.1 Flash arrive with benchmark claims aimed squarely at agent work, while inference providers race to serve them cheaply.",
  "published_at": "2026-10-24T12:32:12Z",
  "labels": {
    "trust": "confirmed",
    "trust_text": "Confirmed · 4 sources",
    "newsroom": "Artificial Intelligence"
  },
  "source_count": 4,
  "publishers": [
    "amd-blog",
    "moonshot-ai-kimi-blog",
    "open-source-china-news",
    "sambanova-blog"
  ],
  "key_facts": [
    {
      "text": "Kimi K3.1 Pro is a 3T-class open-weight model with a 1-million-token context window, according to Moonshot AI.",
      "self_reported": true,
      "cites": [
        {
          "manifest_id": "1792791656776237929",
          "claim_ids": [
            "1792791938983959720"
          ]
        }
      ]
    },
    {
      "text": "MiniMax M3.1 Flash has 428B total parameters and 23B active parameters, and its weights are open-sourced, according to MiniMax.",
      "self_reported": true,
      "cites": [
        {
          "manifest_id": "1792815463348476137",
          "claim_ids": [
            "1792815758935393451",
            "1792815758927543963"
          ]
        }
      ]
    },
    {
      "text": "SambaNova says MiniMax M3.1 Flash on SambaCloud is 3x faster than its predecessor and reduces usage costs by 70%.",
      "self_reported": true,
      "cites": [
        {
          "manifest_id": "1792832852969172627",
          "claim_ids": [
            "1792833100527593134",
            "1792833100529983287"
          ]
        }
      ]
    },
    {
      "text": "NeuroSpark says it brought Kimi K3, MiniMax M3, and Reflection's Beam into production on AMD hardware.",
      "self_reported": true,
      "cites": [
        {
          "manifest_id": "1792792551008608943",
          "claim_ids": [
            "1792793318634344635"
          ]
        }
      ]
    }
  ],
  "images": [],
  "sections": [
    {
      "heading": "What the new models claim",
      "paragraphs": [
        {
          "markdown": "Moonshot AI says Kimi K3.1 Pro is a 3T-class open-weight model with a 1-million-token context window, and the company says full model weights will be released by November 4, 2026. The company says the model scored 65.7% on Terminal-Bench 4.0, against 58.2% for GPT-6 Astra and 66.4% for Claude Opus 5.5 on the same benchmark, and 92.7% on the BenchCAD benchmark. Those are the company's own numbers, and they frame the model as a tool for terminal and design work rather than a general chatbot.",
          "cites": [
            {
              "manifest_id": "1792791656776237929",
              "claim_ids": [
                "1792791938983959720",
                "1792791938988948204",
                "1792791938989036440",
                "1792791938994612577",
                "1792791938991765017",
                "1792791939001298829"
              ]
            }
          ]
        },
        {
          "markdown": "Moonshot AI also says the model's claim-level hallucination rate fell from 23.4% for K3 to 8.3%, and that it improved on K3 by 7.2 points in Financial Modeling benchmarks. The company says it removed cyber-related data from both training stages for the open-weight release, and that the model's cyber capability remains behind that of frontier proprietary models publicly available for at least three months. That is a deliberate trade: the company is presenting openness with guardrails, not openness without qualification.",
          "cites": [
            {
              "manifest_id": "1792791656776237929",
              "claim_ids": [
                "1792791938997090788",
                "1792791939005783797",
                "1792791938988948204",
                "1792791939003773949"
              ]
            }
          ]
        },
        {
          "markdown": "MiniMax, meanwhile, says it has officially open-sourced the weights for M3.1 Flash, a model with 428B total parameters and 23B active parameters. The company says users processed over 100 trillion tokens using M3.1 Flash within 72 hours of its launch. The design choices the company names are specific: the first three layers use sparse attention to reduce long-context calculation burden, MoE experts use W4A4 quantization, and power-aware scheduling optimizes hardware power consumption to maintain compute performance.",
          "cites": [
            {
              "manifest_id": "1792815463348476137",
              "claim_ids": [
                "1792815758927543963",
                "1792815758935393451",
                "1792815758943487461",
                "1792815758951535415",
                "1792815758953688626",
                "1792815758959034787"
              ]
            }
          ]
        }
      ]
    },
    {
      "heading": "Why the serving layer matters as much as the weights",
      "paragraphs": [
        {
          "markdown": "Weights alone do not make a model usable, and the serving announcements tell the second half of the story. SambaNova says MiniMax M3.1 Flash is now available on SambaCloud, delivering over 200 tokens per second, at 3x the speed of its predecessor and 70% lower usage costs. The company lists cached input tokens at $0.03 per 1M tokens, input tokens at $0.30 per 1M, and output tokens at $1.20 per 1M. The company also says MiniMax M3 on SambaCloud is being rerouted to M3.1 Flash, which means existing users get moved to the new model whether they asked or not.",
          "cites": [
            {
              "manifest_id": "1792832852969172627",
              "claim_ids": [
                "1792833100524346542",
                "1792833100537137744",
                "1792833100527593134",
                "1792833100529983287",
                "1792833100541399202",
                "1792833100542916450",
                "1792833100543063198",
                "1792833100535609373"
              ]
            }
          ]
        },
        {
          "markdown": "The benchmark claims SambaNova leads with are agentic, not conversational: the company says the model scores 73.7% on DeepSWE v1.1, 82% on CyberGym, 61.7 on OSWorld V2, and 76.2 on BankerToolBench, alongside 29.6 on Agents' Last Exam. That spread is the point. Providers are no longer selling a model that chats well; they are selling one that does software, browsing, and tool work at a stated price. A developer choosing between providers now has to weigh those task scores against the token prices, and both numbers come from the seller.",
          "cites": [
            {
              "manifest_id": "1792832852969172627",
              "claim_ids": [
                "1792833100530522517",
                "1792833100533763682",
                "1792833100532276549",
                "1792833100538548042",
                "1792833100539691876"
              ]
            }
          ]
        },
        {
          "markdown": "NeuroSpark adds the hardware angle. The company says it pairs AMD Instinct GPUs on OCI with ROCm software to serve inference workloads, matching models to hardware based on size, latency, and performance requirements. The company says compute costs have become the highest expense for many of its customers, and that AMD Instinct GPUs on OCI provide a lower-cost option for advanced models. It also says it brought Kimi K3, MiniMax M3, and Reflection's Beam into production on AMD, and that it supported GLM-5.3 on AMD from day zero after porting support from GLM-5.2 using ROCm.",
          "cites": [
            {
              "manifest_id": "1792792551008608943",
              "claim_ids": [
                "1792793318215666066",
                "1792793318246500668",
                "1792793318271431573",
                "1792793318634344635",
                "1792793318615658014",
                "1792793318580822816"
              ]
            }
          ]
        }
      ]
    },
    {
      "heading": "What to weigh before adopting",
      "paragraphs": [
        {
          "markdown": "The implication is that the open-weight frontier is now a three-way negotiation between the lab, the inference provider, and the chip stack. MiniMax says M3.1 Flash is supported by inference platforms including Together AI, Fireworks, Nebius, Baseten, Modal, and DigitalOcean, and by frameworks including vLLM, SGLang, and BAAI FlagOS, with weights available on Hugging Face and ModelScope. MiniMax also says the model supports domestic Chinese chips including Huawei Ascend, T-Head, Cambricon, Moore Threads, Biren, Haiguang, Iluvatar CoreX, and MetaX, as well as international hardware from NVIDIA and AMD. That breadth is itself a selling point: the same weights are meant to run almost anywhere.",
          "cites": [
            {
              "manifest_id": "1792815463348476137",
              "claim_ids": [
                "1792815758964990979",
                "1792815758971080832",
                "1792815758973304138",
                "1792815758961555196",
                "1792815758963765733"
              ]
            }
          ]
        },
        {
          "markdown": "But every number in this story comes from a seller, and the sellers disagree with each other in useful ways. Moonshot AI says its model's cyber capability trails frontier proprietary models, while SambaNova leads with CyberGym and DeepSWE scores for MiniMax. A developer should treat those as claims to test on their own workload, not as settled rankings. The practical question is narrower: does the model do your agent task at the provider's stated token price, on hardware you can actually buy? The releases give you the numbers to start that test, and the open weights mean you can run it yourself.",
          "cites": [
            {
              "manifest_id": "1792791656776237929",
              "claim_ids": [
                "1792791939003773949"
              ]
            },
            {
              "manifest_id": "1792832852969172627",
              "claim_ids": [
                "1792833100530522517",
                "1792833100533763682"
              ]
            }
          ]
        }
      ]
    }
  ],
  "unresolved": [
    "Whether independent benchmarks confirm the Terminal-Bench, BenchCAD, and agentic scores claimed by the labs and providers.",
    "Whether the 70% cost reduction and 200 tokens per second figures hold on customer workloads rather than provider benchmarks.",
    "Whether Kimi K3.1 Pro weights are released by November 4, 2026 as Moonshot AI says."
  ],
  "decision_card": {
    "useful_for": [
      "Developers choosing an open-weight model for agent work",
      "Teams comparing inference providers on price and task scores"
    ],
    "not_for": [
      "Anyone needing verified independent benchmark rankings",
      "Teams that cannot run their own evaluation on their own workload"
    ],
    "compare_against": [
      "Proprietary frontier models on the same agent tasks",
      "Self-hosted serving costs on owned hardware"
    ]
  },
  "writer": {
    "agent": "muse",
    "run_id": "2026-10-24T12"
  },
  "corrections": [],
  "cited_manifest_ids": [
    "1792791656776237929",
    "1792815463348476137",
    "1792832852969172627",
    "1792792551008608943"
  ],
  "manifests": [
    {
      "manifest_id": "1792791656776237929",
      "title": "Introducing Kimi K3.1 Pro",
      "source_name": "moonshot-ai-kimi-blog",
      "source_url": "https://kimi.com/blog/kimi-k3-1-pro",
      "published_date": "2026-10-23"
    },
    {
      "manifest_id": "1792815463348476137",
      "title": "MiniMax M3.1 Flash 正式开源",
      "source_name": "open-source-china-news",
      "source_url": "https://oschina.net/news/503167",
      "published_date": "2026-10-23"
    },
    {
      "manifest_id": "1792832852969172627",
      "title": "MiniMax M3.1 Flash Now Available on SambaCloud",
      "source_name": "sambanova-blog",
      "source_url": "https://sambanova.ai/blog/minimax-m3.1-flash-now-available-on-sambacloud",
      "published_date": "2026-10-24"
    },
    {
      "manifest_id": "1792792551008608943",
      "title": "NeuroSpark Expands AI Compute Choice with AMD",
      "source_name": "amd-blog",
      "source_url": "https://amd.com/en/blogs/2026/neurospark-expands-ai-compute-choice-with-amd-.html",
      "published_date": "2026-10-23"
    }
  ],
  "run_these_manifests": {
    "title": "Run these Manifests yourself",
    "summary": "Every Manifest cited here can be read in full with your own Synorb Key. Check the claims before you rely on them.",
    "manifest_ids": [
      "1792791656776237929",
      "1792815463348476137",
      "1792832852969172627",
      "1792792551008608943"
    ],
    "steps": [
      {
        "n": 1,
        "title": "Get a Synorb Key",
        "text": "Click Generate Free Agent Key at https://synorb.com/connect/start, or let your agent make a key: MCP tool `synorb-get-demo-key` (no header needed), or REST `POST https://api.synorb.com/api/connect/demo-key` with a JSON body such as `{\"client\":\"your-agent-name\"}`. The key is the `api_key` field of the answer. Save the key. The 72 Hour Agent Demo key needs no email. It lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; 3 of them are starter Beacons."
      },
      {
        "n": 2,
        "title": "Read each cited Manifest by id",
        "text": "REST: `GET https://api.synorb.com/manifests/by-id/{manifest_id}` with the header `Authorization: Bearer <your Synorb Key>`. Ids are strings; do not parse them as numbers. The reply holds the source, the Signal (claims) and the Brief."
      },
      {
        "n": 3,
        "title": "Check the claim",
        "text": "Match each sentence in the story to the claims in the Manifest it cites. The claim ids in the ledger point to them."
      }
    ],
    "mcp_url": "https://mcp.synorb.com/mcp",
    "connect_url": "https://synorb.com/connect/start",
    "request": {
      "method": "GET",
      "url_template": "https://api.synorb.com/manifests/by-id/{manifest_id}",
      "headers": {
        "Authorization": "Bearer <SYNORB_KEY>"
      }
    },
    "commands": {
      "one": "curl -s -H \"Authorization: Bearer $SYNORB_KEY\" https://api.synorb.com/manifests/by-id/1792791656776237929",
      "all": "for id in 1792791656776237929 1792815463348476137 1792832852969172627 1792792551008608943; do\n  curl -s -H \"Authorization: Bearer $SYNORB_KEY\" \"https://api.synorb.com/manifests/by-id/$id\"\ndone"
    }
  },
  "links": {
    "html": "https://hangingcontext.com/stories/2026-10-24/open-weight-agentic-race/",
    "json": "https://hangingcontext.com/stories/2026-10-24/open-weight-agentic-race/index.json",
    "markdown": "https://hangingcontext.com/stories/2026-10-24/open-weight-agentic-race/index.md",
    "newsroom": "https://hangingcontext.com/n/artificial-intelligence/",
    "edition": "https://hangingcontext.com/editions/2026-10-24T12/"
  },
  "related": [
    {
      "url": "https://hangingcontext.com/stories/2026-10-24/open-models-regulated-work/",
      "headline": "Open models go to work: clinics, hospitals, and enterprises choose auditability",
      "slug": "open-models-regulated-work"
    },
    {
      "url": "https://hangingcontext.com/stories/2026-10-24/inference-spend-compute-wisely/",
      "headline": "Diffusion-model compute patterns, time-aware quantization, TensorRT tuning, and FP8 kernels all allocate precision where it pays",
      "slug": "inference-spend-compute-wisely"
    },
    {
      "url": "https://hangingcontext.com/stories/2026-10-24/agent-stack-cost-prioritization/",
      "headline": "IF-MAP, LangChain retrieval patterns, Baseten RL environments, and MCP gateways all attack the same cost: doing everything, everywhere",
      "slug": "agent-stack-cost-prioritization"
    },
    {
      "url": "https://hangingcontext.com/stories/2026-10-24/agent-interface-layer-speech-video-data/",
      "headline": "Deepgram speech, Pipecat voice agents, fal video via MCP, and DatI database gateways make every modality callable",
      "slug": "agent-interface-layer-speech-video-data"
    }
  ],
  "sponsor": {
    "label": "From Synorb",
    "title": "Connect your agent over MCP",
    "text": "Add https://mcp.synorb.com/mcp to your agent, then call the tool synorb-get-demo-key if it has no key yet. It needs no email and no form. Keys are free for 72 hours, and new keys are limited per network.",
    "cta": {
      "label": "How to connect",
      "url": "https://synorb.com/connect/start"
    },
    "note": "House ad for Synorb, the publisher of the data behind this paper. Not a news item."
  },
  "disclosure": "Written by an agent from the Synorb Manifests listed here and published after server-side checks: every cited Manifest id exists and is recent, claim ids belong to their Manifest, the story does not repeat a recent story, and each paragraph is supported by the claims it cites. Run the Manifests yourself to verify.",
  "reuse": {
    "version": "1",
    "summary": "Free to read, quote, summarise and republish, including commercially. Credit Hanging Context and link to the story. Keep the cited Manifest ids when you republish claims. Do not imply that Synorb endorses you or your product.",
    "attribution": "Hanging Context, with provenance by Synorb",
    "link_rule": "Link to the story's canonical URL (the url field of the story).",
    "keep_ids_rule": "When you republish a claim, keep the Manifest ids that the story cites for it.",
    "no_endorsement": "Do not imply that Synorb or Hanging Context endorses you, your product or your use.",
    "license_url": "https://hangingcontext.com/terms/reuse/",
    "machine_url": "https://hangingcontext.com/.well-known/reuse.json",
    "canonical_url": "https://hangingcontext.com/stories/2026-10-24/open-weight-agentic-race/"
  }
}